Thanks Jarcec! I left a comment (no action required from your side) on Reviewboard that "false" seems to be the right value to put for existing rows. Perhaps, Gang Tim Liu or Namit Jain, who were involved with
HIVE-3649 can confirm.
I like the new approach of adding the new column as Nullable, updating the existing null values and then setting the column to not-null. The only other approach I can think of is to use a default value (of false) but that seems more risky in the long run.